View metadata, citation and similar papers at core.ac.uk brought to you by CORE 146 Int'l Conf. Dataprovided Minin byg Covenant | DMIN'16 University Repository|
Common Sense Knowledge, Ontology and Text Mining for Implicit Requirements Onyeka Emebo1,2, Aparna S. Varde1, Olawande Daramola 2 1. Department of Computer Science, Montclair State University, Montclair, NJ, USA. 2. Department of Computer and Information Sciences, Covenant University, Ota, Nigeria [email protected], [email protected], [email protected]
Abstract— The ability of a system to meet its requirements is a intended user. However the very fact that CSK is common, strong determinant of success. Thus effective requirements not all knowledge and requirements that entail common sense, specification is crucial. Explicit Requirements are well-defined will be captured or expressed by the expected user. As Polanyi needs for a system to execute. IMplicit Requirements (IMRs) are describes “We know more than we can tell”. It is therefore assumed needs that a system is expected to fulfill though not the responsibility of the software developer to capture as well elicited during requirements gathering. Studies have shown that as manage the unexpressed requirements in the development a major factor in the failure of software systems is the presence of a suitable and satisfactory system. The application of of unhandled IMRs. Since relevance of IMRs is important for Common Sense Knowledge can improve the identification as efficient system functionality, there are methods developed to well as management of IMRs. Common Sense Knowledge aid the identification and management of IMRs. In this paper, CSK) is defined in [3] as a collection of simple facts about we emphasize that Common Sense Knowledge, in the field of Knowledge Representation in AI, would be useful to people and everyday life, such as "Things fall down, not up", automatically identify and manage IMRs. This paper is aimed and "People eat breakfast in the morning". In [7], the authors at identifying the sources of IMRs and also proposing an describe CSK as a tremendous amount and variety of automated support tool for managing IMRs within an knowledge of default assumptions about the world, which is organizational context. Since this is found to be a present gap in shared by (possibly a group of) people and seems so practice, our work makes a contribution here. We propose a fundamental and obvious that it usually does not explicitly novel approach for identifying and managing IMRs based on appear in people's communications. CSK is mainly combining three core technologies: common sense knowledge, characterized by its implicitness. text mining and ontology. We claim that discovery and handling From the literature, it is observed that a number of reasons of unknown and non-elicited requirements would reduce risks have caused the emergence of implicit requirements some of and costs in software development. which include; i) When a software organization develops a product in a new domain and ii) as a result of knowledge gap Keywords- Implicit Requirements, Common Sense between developers and stakeholders due to the existence of Knowledge, Ontology, Text Mining, Requirement Engineering implicit knowledge. Given this background, we claim that CSK will aid in the I. INTRODUCTION identification of IMRs useful for domain-specific knowledge The challenge of identifying and managing implicit bases. This will be useful for storing domain concepts, requirements has developed to be a crucial subject in the field relations and instances for onward use in domain related of requirements engineering. In [7] it was stated “When processing, knowledge reuse and discovery. Thus we build an critical knowledge, goals, expectations or assumptions of key automated IMR support tool based on our proposed stakeholders remain hidden or unshared then poor framework for managing IMRs using common sense requirements and poor systems are a likely, and costly, knowledge, ontology and text mining. consequence.” With the relevance of implicit requirements The rest of this paper is organized as follows: Section II (IMRs) being identified and related to the efficient presents core technologies. Section III reviews related work functionality of any developed system, there have been on IMRs. Section IV describes our automated IMR process different proposals as well as practical methodologies framework. Section V describes the use and evaluation of the developed to aid the identification and management of IMRs. IMR support tool. Section VI gives the conclusions. Common Sense Knowledge (CSK) is an area that involves making a computer or another machine understand basic facts as intuitively as a human would. It is an area in the realm of II. BACKGROUND Knowledge Representation (KR) which involves paradigms In this section, an overview of the concepts relevant in for adequate depiction of knowledge in Artificial Intelligence CSK, ontology, text mining and natural language processing (AI). The area of CSK is being researched for its use in is presented. This is useful in order to understand our proposed identification and capturing of implicit requirements. IMR framework later. Since AI is aimed at enabling machines solve problems like humans, there is a need for common sense knowledge in A. Common Sense Knowledge AI systems to enhance problem-solving. This not only Common Sense Knowledge (CSK) according to [3] is a involves storing what most people know but also the tremendous amount and variety of knowledge of default application of that knowledge [8]. In software engineering, assumptions about the world, which is shared by people and the development of systems must meet the needs of the seems so obvious that it usually does not explicitly appear in
ISBN: 1-60132-431-6, CSREA Press © Int'l Conf. Data Mining | DMIN'16 | 147
ࢄ is a set whose elements are called ك communications. Some characteristics of CSK as identified in 2. R the literature are as follows: relations. For r = (c1, c2) ϵ R, it is written x Share: A group of people possess and share CSK. r(c1) = c2. x Fundamentality: People have a good understanding 3. Ao is a set of axioms on O. of CSK that they tend to take CSK for granted. In recent times, there is an increased use of ontologies in x Implicitness:People more often don’t mention or software engineering. The use of ontologies has been document CSK explicitly since others also know it. proposed by different researchers’ in. the field of x Large-Scale: CSK is highly diversified and in large requirements engineering and management According to [40] quantity. the increased use can be attributed to the following: (i) they x Open-Domain: CSK is broad in nature covering all facilitate the semantic interoperability and (ii) they facilitate aspects of our daily life rather than a specific domain. machine reasoning. Researchers have so far proposed many x Default: CSK are default assumptions about typical different synergies between software engineering and cases in everyday life, so most of them might not ontologies. In Requirements Engineering (RE), ontology can always be correct. be used for: 1) describing requirements specification Previous work on common sense knowledge includes the documents and 2) to formally represent requirements seminal projects Cyc [9] and WordNet [5], ConceptNet [20], knowledge [10]. Ontology is an important resources of Webchild [31] and the work by [14] and [24]. Cyc has domain knowledge, especially in a specific application compiled complex assertions such as every human has exactly domain. In the management of IMRs, ontology can provides one father and exactly one mother. WordNet has manually shared knowledge which can be useful in the management of organized nouns and adjectives into lexical classes, with IMRs in similar or cross domain knowledge management. By careful distinction between words and word senses. conceptualizing domain knowledge including the identified ConceptNet is probably the largest repository of common implicit requirement, it enables the easy adoption and sense assertions about the world, covering relations such as identification and also management of IMRs. This reduces hasProperty, usedFor, madeOf, motivatedByGoal, etc. enormous costs in requirement development process.in Tandon et al. [14] automatically compiled millions of triples making “explicit specification” it aids the reduction of of the form
ISBN: 1-60132-431-6, CSREA Press © 148 Int'l Conf. Data Mining | DMIN'16 |
motivated computational techniques for analyzing and software process model oriented to the specific needs of the software representing naturally occurring texts [7]. industry in Mexico. The results of this analysis showed that According to [4] it is made up of the following sub areas KMoS-RE seems to be more suitable than process proposed by which are linked to linguistics; i) Morphology ii) Syntax MoProSoft. The KMoS-RE strategy improved the negotiation iii)Semantics iv)Pragmatics process and understanding about the domain and the software functionality requested and, the number of concepts and relationships The core purpose of NLP techniques is to realize human- was greater. KMos-RE strategy reduces the symmetry of ignorance like language processing for a range of tasks or applications between clients and users, and developers which facilitates the [8]. The core NLP models used in this research are part-of- transference and transformation of knowledge and reduces increases speech (POS) tagging and sentence parsers [7]. POS tagging the presence of unambiguous functional requirements. involves marking up the words in a text as corresponding to a Using requirements reuse for discovery and management particular part of speech, based on both its definition, as well of IMRs has been covered by a few studies. A study that draws as its context. In addition, sentence parsers transform text into on an analogy-making approach in managing IMRs is a data structure (called a parse tree), which provides insight presented in [21]. This study proposes a system that uses into the grammatical structure and implied hierarchy of the semantic case-based reasoning for managing IMR. The model input text [7]. of a tool that facilitates the management of IMRs through NLP is used for our purpose in analysis of requirements analogy-based requirements reuse of previously known IMRs statements to gain an understanding of similarities that exist is presented. The system comprises two major components: between requirements and/or identify a potential basis for semantic matching for requirements similarity and analogy- analogy. NLP in combination with ontology enables the based reasoning for fine-grained cross domain reuse. This extraction of useful knowledge from natural language approach ensures the discovery, structured documentation, requirements documents for the early identification and proper prioritization, and evolution of IMR, which is expected management of potential IMRs. to improve the overall success of software development processes. However, as of now, this has not been adopted in a practical form. The work in [25] presents a model for III. RELATED WORK computing similarities between requirements specifications to Different researchers have proposed various ways for promote their analogical reuse. Hence, requirement reuse is identification of IMRs. While some have presented based on the detection on analogies in specifications. This applications, others have presented theoretical and conceptual model is based on the assumption of semantic modeling frameworks and others take on the investigative approach in abstractions including classification generalization and order to get real life views of practicing software engineers, attribution. The semantics of these abstractions enable the requirements engineers and other specialists in the field on the employment of general criteria for detecting analogies practicality of stated theories, ideologies and concepts. between specifications without relying on other special Authors in [15] carried out a two part research aimed at knowledge. Different specification models are supported identifying the impact of tacit and explicit knowledge simultaneously. The similarity model which is relatively transferred during software development projects. The first tolerant to incompleteness of specifications improves as the part involved an inductive, exploratory, qualitative semantic content is enriched and copes well with large scale methodology aimed at validating the tacit knowledge problems. Although the identification of analogies in spectrum in software development projects. This involved requirements is essential, this study does not discuss the unstructured interviews for data collection, and therefore subject for the management of requirements. A method to assessment in a narrative form. The second part of this highlight requirements potentially based on implicit or research involved the development of a conceptual framework implicit-like knowledge is proposed in [2]. The identification of a model that supports future software development projects is made possible by examining the origin of each requirement, in their tacit to explicit knowledge transfers. [23] developed effectively showing the source material that contributes to it. an approach based on a novel combination of grounded theory It is demonstrated that a semantic-level comparison enabling and incident fault trees. It focuses on Security Requirements. technique is appropriate for this purpose. The work helps to As a result of the threat landscape, there are new tacit identify the source of explicit requirements based on tacit-like knowledge which arise. This research proposes an approach knowledge but it does not specifically categorize tacit to discover such unknown-knowns through multi-incident requirement and its management. Also, in MaTREx [22], a analysis. For this research an analysis and investigation brief review and interpretation of the literature on implicit method was used. It involved refining theoretical security knowledge useful for requirement engineering is presented. models so that they remain pertinent in the face of a The authors describe a number of techniques that offer continuously changing threat landscape. analysts the means to reason the effect of implicit knowledge In a study carried out by [30], using a case study, the and improve quality of requirements and their management. Knowledge Management on a Strategy for Requirements The focus of their work is on evolving tools and techniques to Engineering (KMoS-RE) which was designed to face the improve the management of requirements information problem of management of tacit knowledge (in elicitation and through automatic trace recovery; discovering presence of discovery stage) and obtain a set of requirements that fulfill tacit knowledge from tracking of presuppositions, non- the clients’ needs and expectations, was compared to provenance requirements etc. However, MaTREx still deals requirements elicitation process proposed by MoProSoft; a Mexican more with handling implicit knowledge.
ISBN: 1-60132-431-6, CSREA Press © Int'l Conf. Data Mining | DMIN'16 | 149
Previous work on commonsense knowledge includes the developed for specific purpose or those of business rules. The Cyc project [9] with a goal to codify millions of pieces of ontology library is implemented using Java Protégé 4.1 common sense knowledge in machine readable form that ontology API. enable a machine to perform human-like reasoning on such knowledge. Another source is WordNet [5] in which nouns and adjectives are manually organized into lexical classes, furthermore a distinction is made between words and word senses; yet its limitation is that there are no semantic relationships between the nouns and adjectives with the exception of extremely sparse attribute relations. Another prominent collection of commonsense is ConceptNet [20], which consists mainly of crowd sourced information. ConceptNet is the outcome of Open Mind Common Sense (OMCS) [6]. OMCS has distributed this CSK gathering task across general public on the Web. Through the OMCS website, volunteer contributors can enter CSK in natural language or even evaluate CSK entered by others. Given this overview of the related literature, our proposed research stands out due to the fact that it introduces CSK for early identification and management of IMR. Moreover, it also embeds text mining and ontology to develop a support tool for managing IMRs. This is described next.
IV. THE COTIR FRAMEWORK The architectural framework is made up of three core technologies: text mining/NLP, CSK and ontology as presented in Fig. 1. The core system functionalities are depicted as rectangular boxes, while the logic, data and Fig. 1. Proposed COTIR Framework knowledge artefacts that enable core system functionalities are represented using oval boxes. The detailed description of 4) Common Sense Knowledge Base(CSKB) COTIR is given in below. The common sense knowledge bases of WebChild and domain-specific KBs are used for enhanced identification of A. IMR Identification and Extraction IMR for specific domain. The part of the COTIR architecture that deals with 5) Feature Extractor knowledge representation and extraction is described in this The feature extractor heuristic gives the underlying section. assumptions for identifying potential sources of IMR in a 1) Data Preprocessing requirement document. Due to semantic features on which A Software Requirements Specification (SRS) document natural language text exist and by taking into account previous that has been preprocessed serves as input to the framework. work done [11, 13, 16, 19, 28], the following characteristic Preprocessing is a manual procedure that which entails features underline the significant aspects in a piece of text in extraction of boundary sentences from the requirements terms of surface understanding that could possibly make a document and further representing images, figures, tables etc. requirement implicit: in its equivalent textual format for use by the system. • Ambiguity such as structural and lexical ambiguity. 2) NL Processor • Presence of vague words and phrases such as “to a The NL processor component facilitates the processing of great extent”. natural language requirements for the process that enables • Vague imprecise verbs such as “supported”, feature extractor. The core natural language processing “handled”, “processed”, or “rejected” operations implemented in the architecture are i) Sentence • Presence of weak phrases such as “normally”, selection, ii) Tokenization iii) Parts of speech (POS) tagging “generally”. iv) Entity detection v) Parsing. • Incomplete knowledge. The Apache OpenNLP library [27] for natural language processing was used to implement all NLP operations. 3) Ontology Library V. IMR SUPPORT TOOLUSE AND EVALUATION The ontology library and CSKB both make up the The COTIR framework illustrated in Fig. 1 is used to develop knowledge model of our framework. The ontology library a support for the management of implicit requirements based serves as a storehouse for the various domain ontologies on text mining, ontology and common sense knowledge. We (.owl/.rdf). The domain ontologies are those that have been now describe the use of this support tool for managing IMRs,
ISBN: 1-60132-431-6, CSREA Press © 150 Int'l Conf. Data Mining | DMIN'16 |
followed by a snap shot of the tool in Fig. 2 and 3 then its performance evaluation. B. Performance Evaluation Procedure A. Use of the COTIR Tool The evaluation makes use of the following sets of The process of using the COTIR tool developed in this requirements specification: i) Course Management System work is as follows. (CMS) [33], this project was developed for use at the Step 1: Preprocess the source documents to get the University of Twente course management system. The requirements into text file format and devoid of graphics, requirements for CMS describes basic functionality like images and tables. course enrollment, course material and roster upload, student Step 2: Select existing CSKB to be used for the identification grading and e-mails communication. ii) EMbedded of IMR. Monitoring Project [34], the EMMON project is a European Step 3: Import the requirement documents and domain Research and Development (R&D) project. The project ontology into COTIR environment. captures requirements for smart locations and ambient Step 4: Click on the “analyze” function of the tool to allow intelligent environments (smart cities, smart homes, smart the feature extractor identify potential sources of IMR in the public spaces, smart forests, etc.) iii) Tactical Control System requirement document. (TCS) requirements [35], This project was designed for the Step 5: See the potential IMRs that are identified as well as Naval Surface Warfare Center-Dahlgren Division and Joint their recommended possible explicit requirements. Technology Center/System Integration Laboratory, Research Step 6: Seek the expert opinion on the IMRs; the experts Development and Engineering Center, U.S. could approve or disapprove the recommendations by This three requirements specification documents were code adding/removing the recommendations through COTIR. named R1, R2, R3 as shown in the evaluation table II. We use the following metrics to assess the performance of the system. Precision (P), Recall (R), F-measure (F).
் ் ܴൌ ܲൌ ሺ்ାிேሻ ்ାி
ܴܲ ܨൌ ܴܲ Fig. 2. A Demo Snapshot of the COTIR Tool input/Analysis screen In these formulas, TP, TN, FP and FN are as follows. TP (true positives): number of requirements judged by both the expert and tool as being implicit TN (true negative): number of requirements judged by both the expert and tool as not being implicit FP (false positive): number of requirements judged by the tool as implicit and by the expert as not implicit FN (false negatives): number of requirements judged by the tool as not implicit and by the expert as implicit.
A group of subjects were asked to mark implicitness in the requirement document and also use the tool. The Subjects are a group of computing professionals, Fig. 3. A Demo Snapshot of the COTIR Tool output comprising software developers, academics and research students. They were given the following instructions: 1) for For evaluation of the COTIR (Commonsense Ontology each specified requirement, mark each requirement based on and Text-mining of Implicit Requirement) tool developed, its implicit nature (noting that a requirement may contain we conduct an assessment of its performance using more than one form of implicitness). 2) For each requirement requirements specification. The objectives of the evaluation specify the degree of criticality of each implicitness on a scale are as follows: (1) to assess the performance of the tool by of 1 to 5. (1 being least critical to 5 being most critical). human experts, (2) to determine its usefulness as a support Table I shows a sample identification form. The type of tool for implicit requirements management within an implicitness includes i) Ambiguity (A) ii) Incomplete organization. (3) to identify areas of possible improvement in Knowledge (IK) iii) Vagueness (V) iv) Others the tool.
ISBN: 1-60132-431-6, CSREA Press © Int'l Conf. Data Mining | DMIN'16 | 151
C. Results of Performance Evaluation IMR judged by experts that was also retrieved by the tool is Table II shows the recall, precision and F-scores computed good and still consistent with best practices. The average F- for the tool relative to eight experts’ (E1–E8) evaluation. score which is a harmonic mean of Precision and Recall is For a detection tool, the recall value is definitely more 70.3%. Which shows that the result of the tool’s performance important than precision. In the ideal case, the recall should was good. As for the IMRs marked by human evaluators but be 100%, as it would allow to relieve human analysts from missed by the tool, manual examination has shown that they the clerical part of document analysis [36]. For our tool with represent implicit factors where we could not identify explicit an average recall value of about 73.7%, it show that the tool patterns that would allow to automate IMR detection. Our is fit for practical use, as it marks six out of eight IMR observation from the simulation experiment (see Figs. 3.) is detected by humans and is consistent with best practices. The that the performance of the tool also depends significantly on average precision is 68.22% which shows the percentage of the quality of the domain ontology and CSKB.
Table I: Sample Identification Form
Table II: Recall, Precision and F-Score metrics from 8 experts (E1-E8) RequirementsE1E2E3E4E5E6E7E8Average Precision R1 75 75 75 69.23 66.67 83.33 50 91.67 73.24 R2 66.67 58.33 33.33 58.33 83.33 58.33 75 75 63.54 R3 68.75 81.25 56.25 75 43.75 86.67 50 81.25 67.87 Average 68.22 Recall R1 90 90 75 90 80 83.33 75 78.57 82.74 R2 66.67 70 66.67 58.33 76.92 70 69.23 75 69.1 R3 73.33 72.22 64.29 66.67 58.33 81.25 61.54 76.47 69.26 Average 73.7 F-Score R1 81.82 81.82 75 78.26 72.73 83.33 60 84.62 77.2 R2 66.67 63.63 44.44 58.33 80 63.63 72 75 65.46 R3 70.97 76.47 60 70.59 50 83.87 55.17 78.79 68.23 Average 70.3
common sense knowledge, ontology and text mining to VI. CONCLUSIONS propose an automated IMR framework. Another significant In conclusion, the ability to automatically identify and contribution is that a support tool is developed based on the manage IMRs will mitigate many risks that can adversely proposed framework and is helpful in domain-specific affect system architecture design and project cost. This contexts. This work would useful to AI scientists and research work evolves a systematic tool support framework software engineers. Its targeted applications include which uses common sense knowledge that can be integrated providing software requirements for various AI systems, into an organizational Requirement Engineering procedure where common sense is useful in automation. for identifying and managing IMR in systems development process. This is a direct response to problems in the practice ACKNOWLEDGMENTS of many organizations that have not been addressed by This work was conducted when Onyeka Emebo was a existing requirements management tools. Hence, this work Fullbright Scholar at Montclair State University, USA, addresses the problem of identifying IMRs in Requirements visiting from Covenant University, Nigeria. The authors documents and its further management. The novelty of this thank the source of the Scholarship for this funding. work is that integrates three core technologies, namely,
ISBN: 1-60132-431-6, CSREA Press © 152 Int'l Conf. Data Mining | DMIN'16 |
REFERENCES [22] R. Gacitua, B. Nuseibeh, , P. Piwek, , A.N. de Roeck, , M. Rouncefield, , P. Sawyer, , A. Willis, and H. Yang, “Making Tacit Requirements [1] A. Parameswaran,: Capturing Implicit Requirements (02-08-2011), Explicit”, Second International Workshop on Managing Requirements http://alturl.com/emeej Knowledge (MaRK'09) 2009. [2] A. Stone, and P. Sawyer, : Identifying tacit knowledge-based [23] Rashid, A., Naqvi, S. A. A., Ramdhany, R., Edwards, M., Chitchyan, requirements. In Software, IEE Proceedings-, vol. 153, no. 6, pp. 211- R., & Babar, M. A. (2016, May). Discovering unknown known security 218. IET, 2006. requirements. In Proceedings of the 38th International Conference on [3] Zang, Liang-Jun, Cong Cao, Ya-Nan Cao, Yu-Ming Wu, and C. A. O. Software Engineering (pp. 866-876). ACM. Cun-Gen. "A survey of commonsense knowledge acquisition." Journal [24] G.E. Lebani, E. Pianta: Encoding Commonsense Lexical Knowledge of Computer Science and Technology 28, no. 4 (2013): 689-719. into WordNet. Global WordNet Conference 2012. [4] Copestake, Natural Language Processing: Part 1 of Lecture Notes. [25] Spanoudakis, G. (1996). Analogical reuse of requirements Cambridge: Ann Copestake Lecture Note Series, 2003.44 specifications: A computational model. Applied Artificial Intelligence, [5] C.D. Fellbaum, G.A. Miller (Eds.): WordNet: An Electronic Lexical 10(4), 281-305. Database. MIT Press, 1998. [26] Sugumaran, V., & Storey, V. C. (2002). Ontologies for conceptual [6] Singh P, Lin T, Mueller E, Lim G, Perkins T, Zhu W. Open mind modeling: their creation, use, and management. Data & knowledge common sense: Knowledge acquisition from the general public. In engineering, 42(3), 251-271. Proc. Conf. Cooperative Information Systems, Oct.30-Nov.1 2002, [27] Welcome to Apache OpenNLP (24/10/2012) pp.1223-1237. http://opennlp.apache.org/ [7] Choi, F. Y. (2000, April). Advances in domain independent linear text [28] Wilson, W. M., Rosenberg, L. H., & Hyatt, L. E. (1997, May). segmentation. In Proceedings of the 1st North American chapter of the Automated analysis of requirement specifications. In Proceedings of Association for Computational Linguistics conference (pp. 26-33). the 19th international conference on Software engineering (pp. 161- Association for Computational Linguistics. 171). ACM. [8] Chowdhury, G. G. (2003). Natural language processing. Annual [29] Yatskevich, M., & Giunchiglia, F. (2004). Element level semantic review of information science and technology, 37(1), 51-89. matching using WordNet. In Meaning Coordination and Negotiation [9] Douglas B. Lenat: CYC: A Large-Scale Investment in Knowledge Workshop, ISWC Infrastructure. Comm. of the ACM 38(11), pp. 32-38, 1995. [30] Sánchez, K. O., Osollo, J. R., Martínez, L. F., & Rocha, V. M. (2015). [10] Decker, B., Ras, E., Rech, J., Klein, B., & Hoecht, C. (2005, Requirements engineering based on knowledge: a comparative case November). Self-organized reuse of software engineering knowledge study of the KMoS-RE strategy and the DMS process. Revista Facultad supported by semantic wikis. In Proceedings of the Workshop on de Ingeniería, (77), 88-94. Semantic Web Enabled Software Engineering (SWESE) (p. 76). [31] Tandon, N., de Melo, G., Suchanek, F., & Weikum, G. (2014). [11] Fabbrini, F., Fusani, M., Gnesi, S., & Lami, G. (2001, June). An Webchild: Harvesting and organizing commonsense knowledge from automatic quality evaluation for natural language requirements. In the web. In Proceedings of the 7th ACM international conference on Proceedings of the Seventh International Workshop on Requirements Web search and data mining (pp. 523-532). ACM. Engineering: Foundation for Software Quality REFSQ (Vol. 1, pp. 4- [32] Gharehchopogh, F. S., & Khalifelu, Z. A. (2011). Analysis and 5). evaluation of unstructured data: text mining versus natural language [12] Gruninger, M., & Lee, J. (2002). Ontology-applications and design. processing. In Application of Information and Communication Communications of the ACM, 45(2), 39-41. Technologies (AICT), 2011 5th International Conference on (pp. 1-4). [13] Kamsties, E., Berry, D. M., Paech, B., Kamsties, E., Berry, D. M., & IEEE. Paech, B. (2001, July). Detecting ambiguities in requirements [33] Abma, B. J. M. "Evaluation of requirements management tools with documents using inspections. In Proceedings of the first workshop on support for traceability-based change impact analysis." Master's thesis, inspection in software engineering (WISE’01) (pp. 68-80). University of Twente, Enschede (2009). [14] N. Tandon, G. de Melo, G. Weikum: Deriving a Web-Scale Common [34] Software Requirements Specification – EMMON (12-07-2010) Sense Fact Database. AAAI 2011. http://www.artemis-emmon.eu/deliverables/FP7-JU-EMMON-2010- [15] Dreyer, H., Wynn, M. G., & Bown, G. R. (2015). Tacit and Explicit DL-WP7-003-D7.1-software-requirements-specification- Knowledge in Software Development Projects: Towards a Conceptual document.pdf. Framework for Analysis. In eKnow 2015 7th International Conference [35] TCS Software Requirements Specification (02-12-1999) on Information, Process and Knowledge Management (No. A, pp. 49- https://fas.org/irp/program/collect/docs/sss_20.pdf. 52). ThinkMind. [36] Kiyavitskaya, N., Zeni, N., Mich, L., Berry, D.M.: Requirements for [16] Lami, G., Gnesi, S., Fabbrini, F., Fusani, M., & Trentanni, G. (2004). tools for ambiguity identification and measurement in natural language An automatic tool for the analysis of natural language requirements. requirements specifications. Requir. Eng. 13 (2008) 207–239 Informe técnico, CNR Information Science and Technology Institute, Pisa, Italia, Setiembre. [37] Guber, T. (1993). A Translational Approach to Portable Ontologies. Knowledge Acquisition, 5(2), 199-229. [17] Feldman, R., & Sanger, J. (2007). The text mining handbook: advanced approaches in analyzing unstructured data. Cambridge University [38] Borst, W. N. (2002). Construction of engineering ontologies. 1997. Press. Centre of Telematica and Information Technology, University of Tweenty: Enschede, The Netherlands. [18] Maedche, A. (2012). Ontology learning for the semantic web (Vol. 665). Springer Science & Business Media. [39] Guarino, N., Oberle, D., & Staab, S. (2009). What is an ontology?. In Handbook on ontologies (pp. 1-17). Springer Berlin Heidelberg. [19] Meyer, B. (1985). On formalism in specifications. IEEE software, 2(1), 6. [40] Castañeda, V., Ballejos, L., Caliusco, M. L., & Galli, M. R. (2010). The use of ontologies in requirements engineering. Global journal of [20] R. Speer, C. Havasi: Representing General Relational Knowledge in researches in engineering, 10(6). ConceptNet 5. LREC 2012. [41] Kao, A., & Poteet, S. R. (Eds.). (2007). Natural language processing [21] O. Daramola, T. Moser, G. Sindre, and S. Biffl,: Managing Implicit and text mining. Springer Science & Business Media. Requirements Using Semantic Case-Based Reasoning Research Preview. REFSQ 2012, LNCS 7195, pp. 172–178, Springer-Verlag Berlin Heidelberg (2012).
ISBN: 1-60132-431-6, CSREA Press ©